RZ/V2M: Vision AI Platform Reference Design

Overview

The RZ/V2M evaluation board is capable of easily implementing software development such as camera sensor input image processing, low power consumption AI inference, and video streaming.

The system integrates the Sony IMX568 camera sensor, which features a global shutter image sensor.

The RZ/V2M utilizes the DRP-AI implementation of the compact YOLOv2 application, integrated with ISP support package, specifically tailored for the IMX568 camera.

The LiveBench implementation of the Vision AI on RZ/V2M with IMX568, Global Shutter Image Sensor allows customers to evaluate the RZ/V2M capability to process the images captured from the Sony IMX568 with following setup:

By employing an external image rotation system.
The setup validates the built-in global shutter capabilities of the Sony IMX568 camera sensor, with assistance from the RZ/V2M platform.

Capabilities

This LiveBench showcases the following:

The image rotation system regulates rotations ranging from 1 to 100 revolutions per minute (rpm).
Live video streaming with image rotation is facilitated through the utilization of the Sony IMX568 image sensor.
The image captured by the sensor undergoes processing through a Darknet tinyYOLOv2 algorithm for object detection using the RZ/V2M platform.
Users can choose to temporarily pause the execution of the algorithm to assess global shutter capabilities, with the ability to resume it at their convenience.
Even at higher rotational speeds, object detection still occurs using the Sony IMX568 sensor when the algorithm is temporarily paused.

Features

Using the Darknet tinyYOLOv2 algorithm, the object detection algorithm quickly identifies objects.
Even with swift image rotation, the effectiveness of object detection remains consistent.
The speed of image rotation can be adjusted within a range of 1 to 100 revolutions per minute (rpm).